4.1 Pivoting

FMS monitors the diagonal pivots during the factoring process. You may set tolerances based on machine architecture for printing warning messages when a pivot becomes small, or when a pivot is considered zero. You may also provide a subroutine you write to handle exceptions when a zero pivot is encountered.

4.2 Computing Eigenvalues

FMS includes utilities for computing Eigenvalues using subspace iteration.

4.3 Substructuring

This option allows you to interrupt the factoring and solving process at the bottom of the matrix and extract a reduced system of equations. You may use this for design optimization, nonlinear analyais, superelements and solving nonsymmetric systems when only a few of the matrix terms are nonsymmetric. You select this option with the NEQSUB Parameter.

4.4 Nonlinear Reassembly

This option allows you to reuse part of the matrix factors when only some of the matrix terms have changed. You select this option with the LOWASM Parameter.

4.5 Force Symmetry

This option transforms a matrix stored as nonsymmetric into a symmetric matrix by averaging the off-diagonal terms. You select this option with the LUOK Parameter.

4.6 Strassen and Winograd Algorithms

These algorithms reduce the number of floating point operations for BLOCK and SLAB matrices. They are selected by the IALGOR Parameter